On Wed, Sep 12, 2018 at 10:08:58AM -0600, Angus Ainslie wrote: > On 2018-09-11 09:33, Guenter Roeck wrote: > >I cant put my finger on it but this seems wrong. As i said both src > >and sink should never be true at the same time. I also din’t > >understand why turning off src should power off your board. Ultimately > >my concern is that we may be just painting over the real problem, and > >that would be really bad to do with dt properties. > > > > I agree that this doesn't seem like the correct way of solving the problem. > On this HW (Emcraft iMX8M BSB) I think the PTN5110 chip has been connected > correctly so I'm assuming that it is some quirk of the PTN5110. > > I didn't design the HW or the chip. This is a workaround for "quirky" > hardware and there may be others that don't behave exactly as expected. > I wouldn't be that sure about that. It may as well be that the tcpc driver and/or the tcpm driver are doing something wrong when initializing. I didn't really understand the logs you sent out earlier. It looked like the system would loose power if the TCPC_CMD_DISABLE_SRC_VBUS command is sent. That doesn't really make sense to me since it indicates that the chip sources power to the remote, and turning that off should not result in a local loss of power. Note that the chip is supposed to be able to report if it is sourcing vbus and if VBUS is present, in the POWER_STATUS register. Another question is the content of the ROLE_CONTROL register when the system boots, and the DEVICE_CAPABILITIES settings. Overall I suspect that we don't handle startup for your system correctly in the tcpc driver. The ideal solution would be to find a solution which does not require any devicetree properties, but to do that we'll need to get a better understanding about your system's requirements. Guenter
